用用户偏好构建内存，让手机端AI更懂你且更快。
From Volume to Value: Preference-Aligned Memory Construction for On-Device RAG
- 以用户偏好为压缩版个人上下文，贯穿检索流程
- 内存降2404倍，偏好遵循准确率提升18.79个百分点
- 手机端内存<1MB，查询延迟仅5.21-29.35毫秒
随着基于大语言模型的个人AI代理快速发展，将其部署在设备端对保护隐私和提升响应速度至关重要。为应对真实请求中固有的个人化与上下文依赖性，这类代理必须基于设备本地的个人上下文生成内容。然而，在严格内存预算下，核心瓶颈在于如何存储信息以确保检索与用户偏好对齐。本文提出EPIC（高效偏好对齐索引构建），将用户偏好作为紧凑且稳定的个人上下文形式，并贯穿RAG流程。EPIC从原始数据中选择性保留与偏好相关的信息，使检索结果更贴近用户意图。在涵盖对话、辩论、解释和推荐的四个基准上，EPIC将索引内存减少2,404倍，偏好遵循准确率提升18.79个百分点，检索延迟降低32.17倍。在设备端实验中，EPIC内存保持在1MB以下，跨三平台查询延迟为5.21至29.35毫秒，支持偏好漂移下的流式更新。代码与数据已开源。
原文摘要 · Abstract (English)
With the rapid emergence of personal AI agents based on Large Language Models (LLMs), implementing them on-device has become essential for privacy and responsiveness. To handle the inherently personal and context-dependent nature of real-world requests, such agents must ground their generation in device-resident personal context. However, under tight memory budgets, the core bottleneck is what to store so that retrieval remains aligned with the user. We propose EPIC (Efficient Preference-aligned Index Construction), which focuses on user preferences as a compact and stable form of personal context and integrates them throughout the RAG pipeline. EPIC selectively retains preference-relevant information from raw data and aligns retrieval toward preference-aligned contexts. Across four benchmarks covering conversations, debates, explanations, and recommendations, EPIC reduces indexing memory by 2,404 times, improves preference-following accuracy by 18.79 %p, and achieves 32.17 times lower retrieval latency over the best-performing baseline. In on-device experiments, EPIC maintains under 1 MB memory and achieves 5.21 to 29.35 ms/query latency across three platforms, while supporting streaming updates under preference drift. Our code and data are available at https://github.com/UbiquitousAILab/EPIC.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。